The effect of timing of tagging on streamer-tag recapture rates for American lobster (Homarus americanus)

Item Type article

Authors Comeau, Michel; Mallet, Manon

Download date 25/09/2021 19:43:21

Link to Item http://hdl.handle.net/1834/30994 476

Abstract—Streamer tags are com- The effect of timing of tagging on streamer-tag monly used to study the ecology and population biology of the American lob- recapture rates for American lobster ster (Homarus americanus). Aquarium observations suggest that streamer (Homarus americanus) tag loss, either through tag-induced mortality or tag shedding, is related to Michel Comeau the molt stage of the lobster at the time of tagging, and the molting event itself. Manon Mallet Tag-induced mortality, where lobsters Department of Fisheries and Oceans did not molt, occurred within eleven 343 University Ave and sixteen days following tagging for Moncton, New Brunswick, Canada E1C 9B6 lobsters tagged in postmolt (4%) and E-mail-address (for M. Comeau): [email protected] late premolt (10%) stages, respectively; whereas no lobsters tagged in early premolt or intermolt stages died. Tag- induced mortality at time of molting was observed for lobsters tagged in late premolt stage (11%), and tag shed- ding was observed for lobsters tagged Tagging methods to study the move- participation could not explain the dif- both in early (25%) and late premolt ment, growth, and exploitation rate ference between recapture rates for (11%) stages, but was significantly for the American lobster (Homarus lobsters tagged in premolt and postmolt higher (P=0.014) for lobsters tagged americanus) have improved over the stages for a given year; hence possible in early premolt stages. Autopsies last 70 years. One major improvement tag loss at molting was suspected as revealed that lobsters died mainly of in the mid 1960s was the introduction the cause of the lower recapture rates organ perforations (hepato-pancreas of an insertion tag called the “sphyrion (Comeau et al., 1999). Furthermore, in and pericardial sac) following the tag- tag” that is anchored to muscle tissue an attempt to estimate mortality rates, ging process, and rupture of the dorsal (Scarratt and Elson, 1965), instead of Comeau and Mallet (2001) used a thoraco-abdominal membrane during body tags (Templeman, 1935) or cara- mark-recapture model and simulations the molting process. The total tag loss was estimated at 4% for lobsters tagged pace-piercing tags (Wilder, 1953) used to evaluate the best estimator. They after molting, and 27% and 31% for lob- earlier. By the late 1980s, the sphyrion concluded that the level of tag loss is sters tagged in early and late premolt tag was replaced by another insertion high and could be a serious problem for stages, respectively. There was no tag tag called the polyethylene “streamer estimating fi shery parameters for the loss for lobsters tagged in the intermolt tag” (Landsburg, 1991; Moriyasu et al., American lobster if information on tag stage during four months of labora- 1995) initially developed for shrimps loss is not available. tory observations (July−October). To (Penaeus spp.; Marullo et al., 1976). Moriyasu et al. (1995) showed that minimize streamer tag loss, lobsters Insertion tags have the advantage sphyrion tag loss for lobsters held in should be tagged during the intermolt of being retained through a series of aquaria varies between 3% and 23% or postmolt stage. Based on fi eld stud- molts, thus providing information on depending on the molt stage at tag- ies, recapture rates for lobsters tagged in premolt stage are always lower than long-term movement and more accurate ging. They also mentioned that lobsters those of lobsters tagged in postmolt data on growth. tagged with sphyrion tags showed a stage. Furthermore, recapture rates Tag loss could greatly bias the esti- signifi cantly lower return rate (19%) during the second year, for lobsters mate of population characteristics and than lobsters tagged with streamer tags that molt in the year following tagging, fi shery parameters (Ricker, 1975). For (44%) in a recapture study in the fi eld were drastically reduced, and no lobster obtaining population estimates from and suggested a possible lower level of was recaptured after four years at large. mark-recapture data, tag loss gener- tag loss among lobsters tagged with Finally, to account for tag loss during ally refers to the reduction of the initial streamer tags. However, they did not es- the fi rst year at large, a minimal adjust- number of tagged animals by means timate the tag loss for streamer-tagged ment of 24.9% (SD 2.9%) and 4.4% (SD other than fi shing. In a series of tagging lobsters. Recently, Rowe and Haedrich 1.6%) for the recapture rate of lobsters tagged immediately before and after the studies, Comeau et al. (1999) noticed a (2001) showed that the shedding rate molting season, respectively, is recom- constant pattern of lower tag recovery for streamer tags in the field could mended. Adjustments beyond one year rates for lobsters tagged in the premolt reach 18% (40% for molted animals and at large are not recommended for the stage than for lobsters tagged in the 11% for nonmolting animals) after 8−12 American lobster at this time. postmolt stage. They indicated that the months based on double tagging with level of fi shermen participation (recov- a secondary carapace marking. They ery rate) could be a possible cause of also found that streamer tag shedding tag loss because they noticed a steady was not related to sex or size, but they Manuscript approved for publication decline of recapture rates where mul- did not study the level of tag-induced 19 February 2003 by Scientifi c Editor. tiyear tagging studies were conducted. mortality. Manuscript received 4 April 2003 at However, because the same type of Various causes can reduce the initial NMFS Scientifi c Publications Offi ce. streamer tag was used and each lobster number of tagged animals, mainly tag Fish. Bull. 101:476-483 (2003). was handled individually, the fi shermen shedding, tag-induced mortality, and Comeau and Mallet: Effect of timing of tagging on tag recapture rates for Homarus americanus 477

death from natural causes (Beverton and Holt, 1957). In The fi rst experiment began on 23 June 1998 before the the present article, “tag shedding” refers to the physical summer molting season with the tagging of 229 hard-shell detachment of the tag from a lobster and “tag-induced male lobsters ranging between 66 and 78 mm CL. To avoid mortality” refers to the actual death of a lobster caused by unnecessary manipulation of the lobsters, the molt stage the tagging process. at tagging was estimated by the number of days between In the southwestern Gulf of St. Lawrence, lobsters are tagging and molting. Individuals that molted within 30 harvested either in the spring prior to the July−August days of and 30 days after tagging were considered lobsters molting season, or in late summer and early fall (early- tagged in late premolt and early premolt stages (Aiken, August to early October, partially during and shortly after 1980), respectively. The molt stage of lobsters that died dur- molting) (Comeau and Savoie, 2001). The purpose of our ing the experiment was determined by observations of the study was to estimate the level of streamer tag loss for pleopods (Aiken, 1980). A total of 191 male lobsters were the American lobster tagged before and after the summer tagged in premolt stage (56 in early and 135 in late premolt (July−August) molting season by using aquarium observa- stage) and 38 in the intermolt stage (lobsters that did not tions. From the results of our aquarium study, we deter- molt over the entire experiments). Observations of premolt mined adjustments of the recapture rate in relation to the tagged lobsters that molted in June and July ended on 8 molt stage at the time of tagging. September 1998. Observations of the remaining lobsters tagged on 23 June 1998 ended on 30 October 1998. Before releasing the lobsters in the water, their wounds from the Materials and methods tag insertion were examined for infection. The second experiment began 9 September 1998 with the Aquarium observations tagging of 187 soft-shell male lobsters in the postmolt stage based on shell condition criteria (Aiken, 1980; Comeau and Two experiments were carried out at the “Aquarium et centre Savoie, 2001). They ranged in size between 66 and 83 mm marin” (New Brunswick Department of Agriculture, Fisher- CL. Observations of postmolt tagged lobsters ended on 30 ies and Aquaculture) in Shippagan, New Brunswick, with October 1998. lobsters captured in Baie des Chaleurs (47°52′N; 64°52′W). Because the main focus of our study was to investigate tag Field studies loss in relation to the molting stage, only males were con- sidered because they have a higher probability of molting Comparison of recapture rates from fi eld studies Six annually compared to sexually mature females (Comeau tagging studies were carried out after the fi shing season and Savoie, 2001). All lobsters caught were brought to the (May−June) in Caraquet, New Brunswick, between 1994 laboratory where carapace length (CL) and shell rigidity and 1996. Male and female lobsters were captured, tagged, were recorded (the latter with a durometer) (Comeau and and released on the commercial fi shing grounds. For each Savoie, 2001). In both experiments, lobsters were tagged by year, tagging was done before (in early July with hard- the same person to avoid variability in tagging procedure shell lobsters) and after (in mid-September with soft-shell following the technique described by Moriyasu et al. (1995). lobsters) the molting season. In 1995, the July tagging was Streamer tags manufactured by Hallprint Pty. Ltd. (15 delayed one week because the lobster fi shing season ended Crozier Rd, Victor Harbor, South Australia, 5211 Australia) on 7 July instead of 30 June. In 1996, tagging was carried were used. As is routinely done in our tagging studies in out in early October instead of mid-September because of the fi eld (Comeau et al., 1998, 1999), tagged lobsters were bad weather. As with the aquarium experiments, the same kept in a holding tank for a minimum of 30 min following person performed the tagging procedure. Tagged lobsters tagging, and dead lobsters were removed from the experi- were kept in a holding tank for a minimum of 30 min ment. Tagged lobsters were then transferred to large tanks following tagging and any dead lobsters were removed. partitioned with 25 × 25 cm individual compartments. These Finally, an awareness campaign described in Comeau et tanks were supplied with running seawater at ambient al. (1998) was conducted to maximize the participation temperature. Lobsters were fed rainbow smelt (Osmerus of fi shermen in reporting tagged lobsters as tag recovery mordax) twice a week. Lobsters were examined three times took place during the fi shing seasons following the year of a day in July and August, and on a daily basis for the rest tagging. of the experiment. The date of tag shedding, of molting, or of death was recorded for each lobster. Adjustment of the recapture rate Recapture rates ob- Limited aquarium space prevented the use of a control served during the fi shing season following tagging were ad- group of untagged lobsters. However, autopsies were per- justed by using the information from the aquarium obser- formed on all lobsters that died in the course of the experi- vations. In contrast to tag misreporting that biases the ments in order to identify the cause of death. The following estimated number of recaptured lobsters, tag loss affects tagging traumas, causing death, were identifi ed (Krouse the number of tagged animals (N) iinn tthehe ppopulationopulation aavail-vail- and Nutting, 1990): 1) perforation of vital organs, such as able to the fi shery. To account for possible tag loss in our the pericardial sac and the hepato-pancreas; 2) rupture fi eld studies,N wwasas adjustedadjusted aass ffollows.ollows. LLetet p represent the of the thoraco-abdominal membrane; and 3) necrosis or tag-retention rate parameter estimate based on aquarium infection of lobster tissue at both the point of entry and observations with variance p(1−p)n−1, where n is the total exit of the tag. number of animals observed in each aquarium experiment. 478 Fishery Bulletin 101(3)

Table 1 Number and rate (in percentage) of tag shedding and tag induced mortality for American lobsters (Homarus americanus) tagged in postmolt, intermolt, and premolt stages in the aquarium experiments. Lobsters tagged in the premolt stage have been separated into early and late premolt divisions, and n is the number of lobster used to calculate the adjusted recapture rate.

Premolt

Early Late Total Intermolt Postmolt

Initial number1 (n) 56 135 191 38 183 Tag loss without molting Tag-induced mortality 0 13 (10%) 13 (7%) 0 7 (4%) Shedding 0 0 0 0 1 (<1%) Total 0 13 (10%) 13 (7%) 0 8 (4%) Tag loss during or after molting Tag-induced mortality 1 (2%) 14 (10%) 15 (8%) Shedding 14 (25%) 15 (11%) 29 (15%) Total 15 (27%) 29 (21%) 44 (23%) Total tag loss 15 (27%) 42 (31%) 57 (30%) 0 (0%) 8 (4%)

1 Number used in the experiment after accounting for lobsters that died within 30 minutes of tagging.

The adjusted number of tagged animals released during the Secondly, a value for m˜ wwasas obtained from a binomial dis- fi eld studies effectively available to the fi shery is equal to tribution with parameters Ñadjadj aandnd a conditionalconditional recapturerecapture rate tc. A simulated recapture rate was calculated as Nadj = Np, ˜ −1 (1) t = m (Ñadj) . (5) with variance The process was repeated 1000 times and the 90% CI for t˜

V(Nadj) = Np(1 – p ). (2) was defi ned as the 5% and 95% quantiles of the resulting 1000 Monte-Carlo values of t˜. The fi eld study recapture rate (t) is defi ned as

−1 t = m(Nadj) , (3) Results where m = the number of tags returned. Aquarium observations

The variance of t can be calculated by using conditional Tag-induced mortality was observed within 30 min of tag- −1 theory and an approximate variance of (Nadjadj) (Seber, ging for two lobsters tagged in the late premolt stage (1.5%) 1982). In this study, we used Monte-Carlo simulations to and for four in postmolt stage (2.1%) but not for lobsters obtain the 90% confi dence interval (CI) for t. The simula- tagged in early premolt and intermolt stages. The autopsies tions were carried out in two steps to include variability revealed that all of these lobsters died from perforation of associated with tag loss from the initial N animals tagged the pericardial sac and these lobsters were not used in our and released, and variability in recapture rate. First, as- experiments. suming that the number of tags retained (Nadj) follows a Tag loss of premolt lobsters was associated solely with binomial distribution with parameters N and p, a random tag-induced mortality and this was restricted to 10% (13) number (Ñadj) was selected from a Binomial(N,p), where N of the late premolt lobsters (Table 1). Autopsies revealed was set equal to the number of animal tagged during a par- that six lobsters died within three days of tagging from ticular study, and p equal to the proportion of tag retention the rupture of the dorsal thoraco-abdominal membrane, estimated from the aquarium study. With this simulated one died after one day from the perforation of the peri-

Ñadj a conditional value of t was derived as cardial sac, and six died between three to sixteen days after tagging from a perforated hepato-pancreas (Table −1 tc=mo(Ñadj) , (4) 2). In contrast, only 4% (7) of postmolt lobsters died from tag-induced mortality (Table 1). Autopsies revealed that where “~” = the simulated value; and all deaths were associated with the rupture of the dorsal

mo = the observed number of recaptures in a fi eld thoraco-abdominal membrane. Another death was not study. related to tagging (Table 2). Comeau and Mallet: Effect of timing of tagging on tag recapture rates for Homarus americanus 479

Table 2 Number of lobsters (Homarus americanus) that died during the aquarium experiments from different traumas identifi ed by autop- sies. The number of days between tagging and death, and between molting and death, are indicated in parentheses in the “Before molting” and “During or after molting” categories, respectively.

Premolt stage

Early Late Postmolt stage

Total number of lobster in the experiments 56 133 183 Before molting Perforation of the pericardial sac 0 1 (1) 0 Rupture of the dorsal thoraco-abdominal membrane 0 6 (1−3) 7 (1−11)1 Perforation of the hepato-pancreas 0 6 (3–16) 0 No sign of trauma (death not related to tagging) 0 0 1 (5) During or after molting Rupture of the dorsal thoraco-abdominal membrane 0 11 (<1−1) — Poor healing of the entry and exit wounds revealing a punctured dorsal thoraco-abdominal membrane 1 (1) 3 (5−57)2 — No sign of trauma (death not related to tagging) 0 1 (<1) —

1 Five lobsters died within four days, and two more died after nine and eleven days. 2 Two lobsters died fi ve and ten days after molting, and one lobster died 57 days after molting from severe infections at the tag-insertions sites.

A total of 70% (134) of the lobsters tagged in the premolt within 11 days of tagging. The death of the other individual stage molted between 30 June and 8 September (between was not related to tagging for there was no sign of tagging 7 and 77 days after tagging) without any tag loss (Table 1). trauma. Three percent (5) of the lobsters had a misaligned Tag-induced mortality represented 8% (15) of the tag loss tag at the end of the experiment. (Table 1). Autopsies revealed that the majority (11) of the From a total of 229 lobsters initially tagged for the lobsters tagged in late premolt stage died from the rupture fi rst experiment, 17% (38) were in the intermolt stage. of the dorsal thoraco-abdominal membrane (Table 2); the Although the sample number was small, none of these tag was still attached to the old membrane and had caused lobsters shed their tag, died from the tagging procedure, the rupture of the new one. The four remaining lobsters or molted. This fi nding indicates that these small male died from poor healing of the tissue at the point of entry lobsters (67−78 mm CL) skipped an entire molting season and exit of the tag, revealing a punctured dorsal thoraco- (July−August) in 1998 and retained their tags without risk. abdominal membrane (Table 2). For all lobsters that healed Only one (3%) lobster had a misaligned tag at the end of poorly and later died, one side of the tag had been pulled the experiment. No lobster that was returned to the sea inside the body cavity during molting. One death was not after the experiments had infected wounds from the tag related to tagging but was caused by the crusher claw insertion procedure. becoming wedged within the old carapace, thus prevent- ing the animal from successfully completing the molting Field studies process. Furthermore, the autopsy for this lobster did not reveal any sign of tagging trauma. Tag-induced mortality was observed within 30 min in Tag shedding was observed during molting for 21% (29) every tagging study conducted between 1994 and 1996, of the lobsters tagged in the premolt stage (Table 1). Tag varying in rate between 0.4% to 3.5%. The majority of the shedding was observed for lobsters tagged in both early and lobsters tagged in early July 1994, 1995, and 1996 were in late premolt stages (Table 1) but was signifi cantly higher the premolt stage because 94% to 99% had molted before (χ2=5.9; P=0.014) for lobsters tagged in the early premolt being recaptured the following fi shing season. Except for stage. Twenty-one percent (28) of lobsters tagged in the two lobsters (1.1%) tagged in 1994 that were recaptured premolt stage that molted had a misaligned tag (pulled to the following fi shing season with a size increase, all of lob- one side). sters tagged in September−October were in the postmolt Only 4% (8) of the lobsters tagged in the postmolt stage stage. Recapture rates based on the fi rst recapture period died during the second experiment, and only one individual for lobsters tagged in early July (premolt) were signifi - (<1%) shed a tag (Table 1). The autopsies revealed that tags cantly lower than the recapture rates for lobsters tagged were not well embedded in the muscle tissue and large in September−October (postmolt) for the 1994 (χ2=10.6; scars were observed on the dorsal thoraco-abdominal mem- P=0.001) and 1995 (χ2=11.4.; P=0.0007) tagging, but not brane of seven individuals (Table 2). These lobsters died for the 1996 (χ2=2.0; P=0.156) tagging (Table 3). Although 480 Fishery Bulletin 101(3)

Table 3 Numbers and percentage (in parentheses) of lobsters (Homarus americanus) tagged and recaptured during the fi eld tagging studies. Lobsters were tagged before (July) and after (September−October) the molting season between 1994 and 1996, and recaptured the following fi shing seasons (May−June). The recapture rates during the fi rst fi shing season were adjusted on the basis of the tag loss observed in aquarium observations, and Monte-Carlo simulations were used to calculate the confi dence interval. The recapture is based only on the number of legal-size (LS=66.7 mm) lobsters released for the experiment (n). No lobster was recaptured during its fourth year at liberty.

Number of LS Number Number Number Adjusted 90% Dates of lobsters released recaptured recaptured recaptured recapture rate confi dence tagging (n) (1st fi shing season) (2nd fi shing season) (3rd fi shing season) (1st fi shing season) interval

6−7 Jul 1994 476 153 (32.1%) 8 (1.6%) 0 (0.0%) 42.7% (39.0, 48.0) 14−15 Sep 1994 427 182 (42.6%) 10 (2.3%) 4 (0.9%) 44.6% (40.7, 48.9) 11−13 Jul 1995 320 79 (24.7%) 5 (1.6%) 2 (0.6%) 32.9% (28.0, 38.7) 19−20 Sep 1995 442 160 (36.2%) 15 (3.4%) 3 (0.7%) 37.9% (34.3, 42.1) 3−5 Jul 1996 111 31 (27.9%) 3 (2.7%) 0 (0.0%) 37.2% (28.7, 48.1) 2−3 Oct 1996 355 125 (35.2%) 22 (6.2%) 3 (0.8%) 36.8% (32.5, 41.6)

lobsters were recaptured for up to three years following tagging manipulations, the rupture of the dorsal thoraco- their tagging, the recapture rate was drastically reduced abdominal membrane for lobsters tagged in the postmolt between the fi rst and second recapture period (Table 3). No stage might be diffi cult to avoid. Tagging postmolt lobsters lobsters were recaptured after four years. was sometimes diffi cult because the abdominal muscles of Based on the information from our aquarium observa- some lobsters were thin, and insuring the insertion of the tions, the adjusted recapture rates of the fi eld tagging stud- tag through these muscles was a delicate operation. Thus, ies showed no difference between lobsters tagged in the streamer tags not completely embedded into well-devel- premolt and postmolt stages (Table 3). Because the 1995 oped abdominal muscles could be a serious problem for tag July tagging was carried out one week later and the molt- retention. Tag shedding without molting was observed only ing season was early that year (Comeau and Savoie, 2001), for one lobster tagged in the postmolt stage. In general, the adjusted recapture rate, based only on late premolt lobsters tagged with streamer tags during intermolt and lobsters, was 35.2% (CI 30.8, 42.1). The confi dence inter- postmolt stages had a lower level of tag loss than lobsters vals calculated from Monte-Carlo simulations were always tagged in the premolt stage. The level of streamer tag loss wider for lobsters tagged in the premolt stage (Table 3). was the highest for lobsters tagged in the late premolt stage. Tag-induced mortality observed during or shortly after Discussion molting was almost exclusively observed for lobsters tagged in the late premolt stage. Although lobsters tagged Our results suggest that streamer tag loss for the American in the early premolt stage did not die from tagging trauma, lobster depended upon the molt stage of the lobster at the they shed proportionately more tags during molting than time of tagging. Tag-induced mortality without molting was lobsters tagged in the late premolt stage. Unlike tag loss for observed mainly within four days of tagging for lobsters lobsters that did not molt, tag loss at molting was not the tagged in late premolt and postmolt stages; whereas lob- result of the perforation of internal organs or a question- sters tagged in early premolt and intermolt stages seemed able tag insertion during tagging but was caused by the to be less affected by tagging trauma because none died tag being fi rmly attached to the old thoraco-abdomincal following the tagging process. Autopsies revealed that the membrane and being shed, still attached to the exuvia. For length of time between tagging and death was related to a tagged lobsters that died during molting, which were al- specifi c tagging trauma and depended upon the molt stage. most exclusively lobsters tagged in the late premolt stage, Perforation of the pericardial sac caused massive bleed- tag loss was caused by massive bleeding resulting from the ing and resulted in death within 30 min of tagging. The rupture of the entire new thoraco-abdominal membrance rupture of the dorsal thoraco-abdominal membrane, which by the tag that was still fi rmly attached to the old thoraco- also caused bleeding, killed the lobsters within days. Per- abdominal membrane. In addition, we observed a high level foration of the hepato-pancreas killed the lobsters within (21%) of tag misalignment due to the partial attachment weeks. These types of tagging trauma that caused death of the tag to the old carapace for lobsters that did not have were reported by Krouse and Nutting (1990) for American tag loss during molting. For these lobsters, incorrect in- lobsters tagged with Australian western rock lobster (Pan- sertion of the streamer tag was not an issue. Hence, fi rm ulirus longipes cygnus) insertion tags. With more careful tag attachment of the tag to the dorsal thoraco-abdominal tagging manipulations, these types of trauma could pos- membrane of the old carapace (which causes tag-induced sibly be reduced. However, even with extreme care during mortality or tag shedding) is a serious problem. Comeau and Mallet: Effect of timing of tagging on tag recapture rates for Homarus americanus 481

The level of streamer tag shedding in nature seems to be sters tagged immediately before or after the molting sea- higher than that observed in our aquarium observations son. The effi ciency of streamer tags compared to sphyrion and could provide additional information for adjusting tags had already been established (Moriyasu et al., 1995; streamer-tag recapture rates. From a fi eld study, Rowe and Comeau et al., 1999). Moriyasu et al. (1995) reported that Haedrich (2001) observed that 11% of nonmolted lobsters there was a signifi cantly greater recapture rate for lob- shed their tags in a one-year period. If we assumed that sters tagged with streamer tags (44%) compared to those these lobsters were tagged in the intermolt and postmolt tagged with sphyrion tags (19%). Based on the results of stages, their estimate for streamer-tag shedding is higher Comeau et al. (1999), the recapture rate of lobsters tagged than our estimate of less than 1%. Different tagging tech- with sphyrion tags is 22% and 16% for lobsters tagged niques could explain this difference because Rowe and Hae- in premolt and postmolt stages, respectively, compared drich (2001) tagged larger lobsters (>100 mm) differently to 33% and 45% for lobsters tagged with streamer tags. (with the tag inserted in only one abdominal muscle) than These recapture rates corroborate aquarium observations did Moriyasu et al. (1995) for smaller lobsters. The differ- by Moriyasu et al. (1995) on the tag retention of sphyrion ence could also be explained by the artifi cial conditions of tags and ours on the tag retention of streamer tags for our aquarium experiment. Under natural conditions tagged lobster tagged at various molt stages. lobsters could shed their tags through interspecifi c interac- Knowledge of the level of tag loss is paramount for ad- tions (Rowe and Haedrich, 2001), intraspecifi c interactions, justing the recovery rate to estimate population character- and by being dislodged by obstacles in their habitat (Ennis, istics and fi shery parameters for the American lobster. We 1986; Krouse and Nutting, 1990). Streamer tag loss related observed that the recapture rate dropped signifi cantly in to inter- and intraspecifi c interactions and the habitat has the second and third years at large; this fi nding suggests already been reported for the brown shrimp (Howe and a high level of tag loss. A similar multiple-years recapture Hoyt, 1982; P. aztecus) and the tiger prawn (Hill and Was- rate pattern was observed for six other sites within the senberg, 1985; P. esculentus). However, more research would southwestern Gulf of St. Lawrence (Comeau and Savoie, be needed to identify the cause of tag shedding in nature 2002). Rowe and Haedrich (2001) indicated that the and assess its variability in relation to different lobster streamer tag shedding level for lobsters that molted almost habitat before the recapture rate could be adjusted based a year later reached 40%. This high level of tag shedding, on inter- and intraspecifi c interactions and the habitat. probably related to the streamer tag remaining fi rmly The overall level of streamer tag loss compared to sphy- attached to the old dorsal thoraco-abdominal membrane rion tag loss seems to be lower, but also depends upon the during molting, might explain the drastic decrease of tag molt stage of the lobster at tagging and molting. In their recaptures observed between the fi rst and the second tag- study, Moriyasu et al. (1995), suggested that sphyrion tag recapture periods in our fi eld study. We believe that the loss mainly occurs within days after tagging or during adjustment of the recapture rate due to tag loss should molting and is related to the lobster molt stage at tagging. be limited to lobsters recaptured within the fi rst year at We observed lower levels of tag loss compared to those of large and prior to the molting season for lobsters tagged Moriyasu et al. (1995), except for the tag shedding during in intermolt and postmolt stages. molting for early premolt lobsters and tag-induced mor- The multiple-years recapture pattern of a high recapture tality for lobsters tagged in late premolt. They observed rate within the fi rst year at large followed by low recapture 3% and 11% of tag shedding without molting for lobsters rates in subsequent years could have a signifi cant impact tagged in early and late premolt stages, respectively, com- on multiyear tagging models. These models that were pro- pared to none in our study. Furthermore, the most striking posed by Seber (1970) could be used to estimate population difference is the level of tag loss that reached 10% and characteristics and fi shery parameters if underlying as- 30% for lobsters tagged in intermolt and postmolt stages sumptions are followed. Based on these multiyear models compared to 0% and <1%, respectively, in our study. The originally developed for birds (Seber, 1970; Brownie et al., difference in tag loss for lobsters tagged in the postmolt 1985), a suite of models adapted for fi shery data and ad- stage could be explained by the physical nature of the justments, mainly by reparameterization, were proposed tags themselves and the tagging techniques. Compared to to address underlying assumption violations (Pollock et the streamer tag that is threaded through two abdominal al., 1991, 2001; Hearn et al., 1998, 1999; Hoenig et al., muscles, the sphyrion tag is anchored to only one muscle by 1998a, 1998b; Frusher and Hoenig, 2001; Latour et al., means of a hypodermic needle (Moriyasu et al., 1995). Be- 2001a, 2001b). Some of these models were developed to cause the muscles of postmolt lobsters (in the early stages) take into account fi shing effort, incomplete mixing, and are not well formed, it is diffi cult to fi rmly embed an object, tag recovery rate. The latter is a composite parameter such as a tag, and the probability of tag loss for a tag em- involving tag retention and tag-induced mortality (tag bedded into only one thin muscle is greater than that for loss), exploitation rate, and tag reporting rate. There is no a tag treaded through two muscles. Hence, it seems that argument that the participation of fi shermen in returning streamer tags are more effective in terms of tag retention tags (tag reporting rate) is very important; however, for compared to sphyrion tags for lobsters tagged in intermolt crustacean fi sheries, underlying assumptions dealing with and postmolt stages, but equally so for lobsters tagged in tag loss (see assumptions 2 and 3 in Pollock, 1991, 2001) the premolt stage. are equally important and have to be addressed. In general, In fi eld tagging studies, streamer tags yielded a good re- Pollock et al. (2001) indicated that the assumptions of no capture rate within the fi rst year following tagging for lob- tag loss could be violated in two ways: by tag loss in the fi rst 482 Fishery Bulletin 101(3)

few days after tagging or by chronic tag loss spread over reviewing the manuscript, and three anonymous reviews an extended period of time, the latter being more diffi cult for thoughtful suggestions that improved the quality of to model. Furthermore, Seber (1970) mentioned that the this manuscript. usefulness of the multiyear models depends not only on the validity of the underlying assumptions but also on the number of recaptures (i.e. parameter estimates based on a Literature cited small number of tag recaptures would be biased). From our aquarium observations and fi eld studies that show a small Aiken, D. E. number of lobsters caught during the second and third re- 1980. Molting and growth. In The biology and manage- capture periods, we conclude that a signifi cant chronic tag ment of lobsters. Vol. I: Physiology and behavior (J. S. Cobb loss does occur for the American lobster due to molting (i.e. and B. F. Phillips, eds.), p. 91−163. Academic Press, New the molt stage of the lobster at tagging and molting itself). York, NY. Chronic tag loss impedes the effectiveness of multiyear re- Beverton, R. J. H., and S. H. Holt. capture models currently used (Hearn et al, 1998; Hoenig 1957. On the dynamics of exploited fi sh populations. U.K. Minist. Agric. Fish., Fish. Invest. 19:1−533. et al., 1998a; Frusher and Hoenig, 2001) because it is not Brownie, C., D. R. Anderson, K. P. Burhnam, and D. S. Robson. taken into account. We believe that assuming only a con- 1985. Statistical inference from band recovery data: a stant short-term tag loss for lobsters tagged with streamer handbook, 2nd ed. U.S. Fish Wildl. Serv. Resour. Publ. tags is inadequate and can only bias estimates of survival 156, 305 p. and exploitation rate. Correcting for chronic tag loss after Comeau, M., W. Landsburg, M. Lanteigne, M. Mallet, P. Mallet, the fi rst year at large for the American lobster, however, G. Robichaud, and F. Savoie. requires further knowledge, and more studies would be 1998. Lobster (Homarus americanus) tagging project in required to fully understand long-term tag loss. Caraquet (1993)—tag return from 1994 to 1997. Can. In conclusion, a high level of streamer tag loss is a major Tech. Rep. Fish. Aquat. Sci. 2216, 35 p. obstacle for using tagging studies to estimate natural mor- Comeau, M., M. Lanteigne, G. Robichaud, and F. Savoie. 1999. Lobster (Homarus americanus) movement in the tality or to apply multiyear models for the American lobster. southern Gulf of St. Lawrence—summary sheets of tagging Because streamer tag loss is related to molting, adjustment projects conducted between 1980 and 1997. Can. Ind. Rep. is diffi cult because the molting frequency is size, sex, and Fish. Aquat. Sci. 249, 111 p. environment dependent (Comeau and Savoie, 2001). In our Comeau, M., and M. Mallet. attempt to estimate mortality at molt, it was found that 2001. Estimating mortality rates by capture-recapture, differences in recapture rates of lobsters tagged in premolt catch-effort and change-in-ratio models for a spring Ameri- and postmolt stages for a given molting period were not sta- can lobster (Homarus americanus) fi shery (LFA 23). Can. tistically signifi cant, thus suggesting a low level of natural Tech. Rep. Fish. Aquat. Sci. 2373, 20 p. mortality during the molt. The recapture rate for the 1996 Comeau, M., and F. Savoie. tagging, for instance, was even higher for lobsters tagged in 2001. Growth increment and molt frequency of the American lobster (Homarus americanus) in the southwestern Gulf of the premolt stage. Hence, tagging with streamer tags to es- St. Lawrence. J. Crust. Biol. 21:923−936. tablish the level of natural mortality during the molt, or any 2002. Movement of American lobster (Homarus americanus) other mortality that could be low, for the American lobster in the southwestern Gulf of St. Lawrence. Fish. Bull. 100: is not recommended. The alternative would be to develop 181−192. another insertion tag with better retention through the Ennis, G. P. molting process. Nevertheless, the streamer tag remains an 1986. Sphyrion tag loss from the American lobster Homarus adequate choice for studying lobster ecology and population americanus. Trans. Am. Fish. Soc. 115:914−917. biology. Streamer tags could be used to tag intermolt and Frusher, S. D., and J. M. Hoenig. postmolt lobsters during single recapture tagging studies 2001. Estimating natural and fi shing mortality and tag to estimate the exploitation rate (Xiao et al., 1999). Based reporting rate of southern rock lobster (Jasus edwardsii) from a multiyear tagging model. Can. J. Fish. Aquat. Sci. on our observations, a minimum adjustment of 24.9% (SD 58:2490−2501. 2.9%) and 4.4% (SD 1.6%) is suggested for lobsters tagged Hearn, W. S., K. M. Pollock, and E. N. Brooks. in premolt and inter- or postmolt stages, respectively, and 1998. Pre- and post-season tagging models: estimation of recaptured during the fi rst recovery period. reporting rate and fi shing and natural mortality rates. Can. J. Fish. Aquat. Sci. 55:199−205. Hearn, W. S., T. Polacheck, K. H. Pollock, and W. Whitelaw. Acknowledgments 1999. Estimation of tag reporting rates in age-structured multicomponent fisheries where one component has The authors wish to thank all fi shermen from Caraquet observers. Can. J. Fish. Aquat. Sci. 56:1255−1265. and the adjacent wharves that returned lobster tags and T. Hoenig, J. M., N. J. Barrowman, W. S. Hearn, and K. H. Pollock. 1998a. Multiyear tagging studies incorporating fishing Brideau, B. Comeau, J. Roussel, and F. Savoie for their tech- effort data. Can. J. Fish. Aquat. Sci. 55:1466−1476. nical assistance in the fi eld and during the tag collection. Hoenig, J. M., N. J. Barrowman, K. H. Pollock, E. N. Brooks, and We especially thank A. Godin and his staff at the Aquarium W. S. Hearn. et centre marin in Shippagan, New Brunswick, for their 1998b. Models for tagging data that allow for incomplete professional help during the aquarium experiment. We also mixing of newly tagged animals. Can. J. Fish. Aquat. Sci. want to thank J. M. Hanson and M. Moriyasu for critically 55:1477−1483. Comeau and Mallet: Effect of timing of tagging on tag recapture rates for Homarus americanus 483

Hill, B. J., and T. J. Wassenberg. Pollock, K. H., J. M. Hoenig, W. S. Hearn, and B. Calingaert. 1985. A laboratory study of the effect of streamer tags 2001. Tag reporting estimation: an evaluation of the reward on mortality, growth, moulting and duration of noctur- tagging method. N. Am. J. Fish. Manage. 21:521−532. nal emergence of the tiger prawn Penaeus esculentus Pollock, K.H., J. M. Hoenig, and C. M. Jones. (Haswell). Fish. Res. 3:223−235. 1991. Estimating of fi shing and natural mortality when Howe, N. R., and P. R. Hoyt. a tagging study is combined with a creel survey or port 1982. Mortality of juvenile brown shrimp Penaeus aztecus sampling. Am. Fish. Soc. Symp. 12:423−434. associated with streamer tags. Trans. Am. Fish. Soc. 111: Ricker, W. E. 317−325. 1975. Computation and interpretation of biological statistics Krouse, J. S., and G. E. Nutting. of fi sh populations. Bull. Res. Board Can. 191, 382 p. 1990. Effectiveness of the Australian western rock lobster Rowe, S., and R. L. Haedrich. tag for marking juvenile American lobsters along the Maine 2001. Streamer tag loss from American lobsters. Trans. coast. Am. Fish. Soc. Symp. 7:94−100. Am. Fish. Soc. 130:516−518. Landsburg, A. W. Scarratt, D. J., and P. F. Elson. 1991. A fi eld comparison of recapture rates of polyethylene 1965. Preliminary trials of a tag for salmon and lobsters. J. streamer and modifi ed sphyrion tags through molting of Fish. Res. Board Can. 22:421−423. lobster (Homarus americanus). J. Shellfi sh. Res. 10, 225 Seber, G. A. F. p. 1970. Estimating time-specific survival and reporting Latour, R. J., J. M. Hoenig, J. E. Olney, and K. H. Pollock. rates for adult birds from band returns. Biometrika 57: 2001a. Diagnostics for multiyear tagging models with appli- 313−318 cation to Atlantic striped bass (Morone saxatilis). Can. J. 1982. The estimation of animal abundance and related Fish. Aquat. Sci. 58:1716−1726. parameters, 2nd ed., 654 p. Griffi n, London. 2001b. A simple test for nonmixing in multi-year tagging Templeman, W. studies: application to striped bass tagged in the Rappahan- 1935. Lobster tagging in the Gulf of St. Lawrence. J. Biol. nock River. Trans. Am. Fish. Soc. 130:848−856. Board Can. 1:269−278. Marullo, F., D. A. Emiiani, C. W. Caillouet, and S. H. Clark. Wilder, D. G. 1976. A vinyl tag for shrimp (Panaeus spp.). Trans. Am. 1953. The growth of the American lobster (Homarus Fish. Soc. 105:658−663. americanus). J. Fish. Res. Board Can. 10:371−403. Moriyasu M., W. Landsburg, and G.Y. Conan. Xiao, Y., J. D. Stevens, and G. J. West. 1995. Sphyrion tag shedding and tag induced mortality 1999. Estimation of fi shing and natural mortalities from tag of the American lobster, Homarus americanus H. Milne experiments with exact or grouped times at liberty. Can. Edwards, 1837 (Decapoda, Nephropidae). Crustaceana J. Fish. Aquat. Sci. 56:868−874. 68:184−192.